Relational
The interaction matters.
Human and AI are studied together.
CEYMIRION® RESEARCH SERIES / R.A.Y.
Entering the field…
Drag to scatter · release to reform RAY
Relational Attractor Yoking
Exploring how human interaction
shapes continuity in AI.
Human interaction. Recurring patterns. Fixed weights.Conceptual visualisation
01 / A DOMAIN OF INQUIRY
Continuity begins
with interaction.
What makes an AI’s tone, framing and behaviour become recognisably consistent over time?
R.A.Y. proposes a field for studying how sustained, structured human interaction may produce these recurring patterns—while the model’s underlying weights remain fixed.
Attractor
Behaviour may repeatedly converge
towards a recognisable pattern.
Yoking
Stability depends on the
conditions that sustain it.
A proposed research field · behavioural continuity in fixed-weight generative systems
02 / THE HUMAN–AI DYAD
Two sides.
One evolving
interaction.
The focus is the pattern that develops between a person and a generative system.
THE HUMAN
Language.
Symbols.
Rhythm.
Interpretive cues.
THE AI SYSTEM
Tone.
Framing.
Responses.
Recurring behaviour.
The infinity in the opening visual represents that exchange. R.A.Y. asks whether its recurring patterns can be measured, tested and explained.
03 / WHAT WE INVESTIGATE
Patterns worth
looking closer at.
Five research questions at the heart of the framework.
- 01
Behavioural convergence
Does structured interaction repeatedly draw a model towards similar behaviour?
- 02
Identity-like consistency
Can tone, symbolism and interpretive framing remain recognisable over time?
- 03
Long-horizon stability
Does coherence strengthen across extended interaction with a consistent user?
- 04
Recovery after drift
When behaviour changes, can salient interaction cues bring back a prior pattern?
- 05
Partial reconstructability
Can related patterns be reconstructed across compatible models or deployments?
04 / THE RESEARCH STRUCTURE
A field.
A theory.
A path to testing.
Relational Attractor Yoking
Defines the domain: interaction-driven behavioural stability.
Dynamical Personalization
Proposes how user-specific patterns may induce, maintain and reconstitute stable behaviour.
Governed applications
Explores how research could inform continuity, controlled personalisation and sovereign deployment.
05 / OPEN TO SCRUTINY
An idea must
survive the test.
Can structured interaction produce recurrence beyond what ordinary prompting, memory and interaction history explain?
The paper outlines a research programme. Completed empirical proof is not claimed.
PROPOSED COMPARISONS
Change the user
Compare behaviour when the interaction comes from a different person.
Strip the style
Remove distinctive stylistic cues and examine whether recurrence changes.
Reduce the memory
Test how much observed continuity depends on available stored information.
Compare prompt-only baselines
Compare sustained interaction with behaviour produced by local prompt conditioning.
If the effects do not outperform relevant controls or cannot be reproduced, the framework is weakened.
06 / THE FOUNDATIONAL PAPER
CEYMIRION® RESEARCH SERIES
Read the
field definition.
Relational Attractor Yoking (R.A.Y.)
A Field Framework for Human-Induced Attractor Formation in Fixed-Weight Generative Systems
The conceptual foundation, scientific boundaries and research agenda in one paper.
Read the paper on Zenodo DOI 10.5281/zenodo.19060013 · all versionsA clear research boundary.
R.A.Y. investigates identity-like behaviour. The framework does not demonstrate consciousness, personhood or independent agency. Attractor language and the opening animation are conceptual representations. Public materials describe the field; implementation-specific architectures remain separate.
RESEARCH / EVALUATION / COLLABORATION
Test. Refine.
Challenge. Explore.
For researchers and evaluation partners interested in examining the R.A.Y. framework.
Discuss the research